Open Stack

In my openstack deployment, I used containers for different openstack components. Since the kernel is shared, some of the global parameters needs to be adjusted, for example, the max open file descriptor. Or you will soon find yourself running out of file descriptors in your containers. This kind of optimization may take time to shape. Also, other than those container advantages, there is no way to do live-migration at the current state of art.
  For the purpose of the Hackathon, I don’t think live migration is needed for most of the cloud application. They just need to have a lot of containers to scale out their cloud application. Also, for those simple application that does not use a lot of file descriptors, they may not hit the issue of using up file descriptors like the mysql process or rabbitmq-server does. But, just in case, some of the cloud applications want to show how to optimized the energy usage of the compute node by moving the VMs around, you could not do that with containers.

Regards,
EJ
F Division, ICL, ITRI


====================================================================
本信件可能包含工研院機密資訊，非指定之收件者，請勿使用或揭露本信件內容，並請銷毀此信件。 
This email may contain confidential information. Please do not use or disclose it in any way and delete it if you are not the intended recipient.